View Single Post
  #2  
Old April 14th 08, 01:41 AM posted to alt.comp.hardware
Paul
external usenet poster
 
Posts: 13,364
Default Memory Issue or Test issue??

bill wrote:
I'll try to make this short, but need to give all the details,
so......

i have basically a gaming system for my wife. Built it for xmas and
because i went high end with video card, i only got 2 gigs RAM. For
these few months, system has been ROCK SOLID, never a crash, fast as
fast......everything great.

Found RAM, same type i have in, had rebate going so i picked up
another 2 gigs.Shortly after install, system would kick out of WOW and
back to desktop, giving memory error. You could always relaunch
without reboot and launch WOW again till it happened again. Pulled
new 2 gigs and system went back to flawless operations.

Downloaded free memory tester (http://www.memtest86.com/) and did some
testing. I decided to do a baseline test using only known good 2 gigs
ram. Test fails hard and locks up system.Move 64k block test. And this
was with first 2 gigs ram that works without errors.Swapped ram with
new, was able to quickly isolate one defective chip, other seems good.
NOW back to old ram that works flawless, installing both sticks, tests
fails as stated above.BUT, if i take system down to 1, 1 gig stick and
test one at a time, both sticks pass, without a problem, multiple
tests. IF i move one of these sticks to memory slot 1 instead of zero,
they still test fine. Only when both sticks are in does memory test
fail.

And now as i writing, wife informs me my flawless system just rebooted
on it own, FIRST failure every. This is 1st ram, so i just swapped
slots, but not ram. Moved ram in slot 0 to slot 1 and vise visa. SEEMS
to be doing fine again, not reboots for over 3 hours.

someone got some idea's??


Each generation of memory, is supposed to be improving on the previous
generation, in terms of the memory bus and sensitivity to loading.

Some of the original DDR2 tests, with two sticks versus four sticks of
memory, showed that only a very slight change of timing was necessary,
to get the same level of stability. Something along the lines, that
if two sticks ran 5-5-5-15, four sticks might run 5-6-5-15. The first
number is CAS, and has the largest effect on performance, but CAS manages
to stay the same.

DDR2 chips have ODT or "on-die termination". Those terminations give the
hardware more options for properly terminating bus signals. There are
other tricks as well, for improving signalling.

Now, that being said, later motherboards have come out, where one slot of
the two on the channel, seems to work better. I don't know why this would be
the case, but there is that possibility.

Memory is now dirt cheap, and with that dirt cheapness comes a need to
make money. About the only thing you can cut corners on, is test time.
I'm sure there are some memory products out there, that aren't tested
at all (solder and ship).

Possible reasons for your problems:

1) Vdimm not high enough for error free operation, at the speed you've
selected for the memory bus. 1/8V is nominal. Products usually mention
a max or suggested value, like 2.0V or 2.1V etc.
2) Accidental overclock of memory, while overclocking processor. Use CPUZ
from cpuid.com to verify that the settings are correct.
3) Overheating Northbridge. Overheating processor. Use Speedfan or the
manufacturer provided hardware monitor utility, to check what sensors
have been provided. Stick a finger on your heatsinks (including chipset).
If you burn yourself on anything, fix it.

You can also check the Newegg reviews for the product you bought, to get
some idea what results other people are getting with the product in
question.

For further help, state motherboard brand/model number, CPU type, memory
brand and model number, CPU and memory speeds (3.0GHz core, FSB1333,
DDR2-800 running at DDR2-800 5-5-5-15 etc). You should be able to get some
of that info from CPUZ.

Paul